AMD revenue and earnings beat Wall Street expectations, announces $6 billion stock buyback, shares rise 4%

Advanced Micro Devices Inc (Nasdaq ticker AMD) released its first-quarter financial results. According to the Benzinga Pro report, AMD reported revenue of $7.44 billion, exceeding analysts' expectations of $7.13 billion, with earnings per share of 96 cents, surpassing the expected 94 cents, representing a 36% year-over-year revenue increase.

Chairman and CEO Lisa Su stated that AMD's core business showed strong growth, with expanded data center and artificial intelligence development, providing an excellent start to 2025, with four consecutive quarters of year-over-year growth. The first quarter's performance and second quarter outlook highlight AMD's product advantages, laying a solid foundation for strong growth in 2025.

AMD's board approved a $6 billion stock buyback program, supplementing the existing $4 billion buyback, with the stock rising 4% following the announcement.

On Tuesday, AMD announced an investment agreement potentially worth $10 billion to provide chip support for an AI company called Humain in Saudi Arabia.

Table of Contents

Toggle

AMD Q1 Financial Report Highlights

• Data Center Revenue: $3.7 billion, growth of 57%
• Client Revenue: $2.3 billion, growth of 68%
• Gaming Revenue: $647 million, decline of 30%
• Embedded Revenue: $823 million, decline of 3%

AMD's Q1 gross margin reached 54%, with operating income of $1.8 billion, and end-of-period cash, cash equivalents, and restricted cash of approximately $6.06 billion.

AMD as Nvidia's Direct Competitor

AMD and Nvidia have long been direct competitors in the AI chip industry. AMD stated it will provide graphics processors for AI and central processors needed to build AI servers to Humain in Saudi Arabia, with Humain also purchasing Nvidia processors.

AMD and Humain Collaborate on Supercomputing AI Center

According to AMD's official press release, AMD and Humain will invest and deploy up to $10 billion, 500 megawatts of AI computing centers over the next five years. The AI Superstructure jointly built by AMD and HUMAIN will be optimized for enterprise, startup, and sovereign AI workloads. HUMAIN will oversee end-to-end delivery, hyperscale data centers, sustainable power systems, and global fiber interconnection, while AMD will provide its full portfolio of AMD AI computing products and AMD ROCm™ open software ecosystem.

With AMD's impressive financial report, stock buyback plans, and news of collaboration with Humain AI in Saudi Arabia, Wall Street analysts are optimistic about AMD's development. Bank of America analyst Vivek Arya raised AMD's target price by $10 to $130 per share. Pure market observation, not investment advice.

Risk Warning

Cryptocurrency investment carries high risk, with potentially volatile prices, and you may lose all your principal. Please carefully assess the risks.

What is Psyche? Three Core Technologies Reshape AI Training Logic

Psyche, as a groundbreaking decentralized AI training network developed by Nous Research, aims to enable global users, whether individuals or small communities, to contribute their GPU resources to collaboratively train large AI models. This breaks the past monopoly of AI training resources concentrated among a few tech giants, ushering in a true era of technology sharing and democratization.

(G7 Competition Regulatory Summit: AI Giants, Please Give Space to Startups, Don't Take Creators' Jobs)

The official blog points out that Psyche is equipped with the following key technologies:

• DisTrO Optimizer: Reduces cross-node data transfer by tens of thousands of times, enabling distributed training even with home GPUs like NVIDIA 3090.

• Blockchain Coordination Mechanism: Core training status and node tasks are managed by a "coordinator" smart contract deployed on Solana, ensuring transparent, fair, and trustless training.

• Node Certification and Rewards: Node contribution records are stored on Solana and tokens are distributed based on performance, incentivizing computing resource participation.

How Does Psyche Achieve True Decentralization?

Psyche's overall architecture is divided into three main roles to ensure efficient, stable, and continuous network training:

• Coordinator: Synchronizes tasks and status on Solana.

• Client: Globally distributed GPU nodes responsible for model training and verification.

• Data Provider: Uploads data for training use.

Language Model "Consilience" Officially in Training

Currently, the Psyche testnet is pre-training a language model called "Consilience" with 4 billion parameters, using up to 20 trillion text data, adopting a Multi-Head Latent Attention (MLA) architecture that demonstrates surprising computational and creative potential despite its modest size.

Psyche Practices AI Democratization: Supporting Academic, Educational, and Medical Applications

The launch of Psyche marks a key milestone in AI democratization, reducing technological monopoly and enabling fair innovation:

• Lowering Technical Barriers: Developers, educational institutions, academic researchers, and medical teams will no longer be limited by expensive hardware or resource constraints, and can participate in advanced AI research.

• Improving Resource Efficiency: Transforming globally idle GPUs into training engines, saving energy and reducing dependence on data centers.

• Ensuring Fairness and Diversity: The decentralized architecture prevents human manipulation, ensuring technology serves all humanity.

(Latest on Google Monopoly Case: U.S. Department of Justice Demands Multiple Anti-Centralization Measures, Including Sharing AI Training Data)

Technology Should Not Belong to the Few, But Should Benefit All Humanity

Nous finally states: "The Psyche testnet is just the beginning."

We anticipate that in the future, more unauthorized nodes will join, along with more model architectures and training methods (such as reinforcement learning and fine-tuning), establishing a global collaborative AI development platform and continuously promoting trustless incentive and comprehensive recording mechanisms through Solana.

Psyche is not just a technological innovation, but a transformation of AI development philosophy, symbolizing a new value: "Opening powerful AI tools to participants worldwide and collectively building a fairer and more innovative future."

As this project gradually matures, we may be witnessing the birth of an era where artificial intelligence truly belongs to everyone.